RCaller is a software library which is developed to simplify calling R from Java. Despite it is not the most efficient way of calling R codes from Java, it is very simple to use and its learning curve is steep. It successfully simplifies and wraps type conversations and makes variables in each languages accessible between platforms. With the calling sequential commands facility, the performance is not lost through a single external process. Although R is single-threaded, multiple R processes can be created and handled by multiple RCaller instances in Java. A Servlet based application can instantiate many RCaller objects as well as it can use a single object by using sequential command invocation ability. The former use multiple environments which do not share the same variable pool, whereas, the latter shares a mutual variable pool and clients can communicate as well. RCaller is written purely in Java and it does not depend on any external libraries, that is, it is ready to run in any machines that Java and R installed. Simulation studies show that the other libraries such as Rserve and rJava outperform the RCaller by means of interaction times. As a result of this, RCaller is not suitable for the projects which have many clients that request relatively single and small computations. [RCaller: A Software Library for Calling R from Java - M. Hakan Satman]
RCaller is a production ready library for calling R functions from within Java. If you are developing Java software that needs enhanced statistical calculations, RCaller is a practical solution to integrate these languages. R has many well-tested and matured packages for automatic time series model selection, clustering, segmentation and classification, non-linear and robust regression estimations, data and text mining, linear and non-linear programming, generating plots, function optimization besides other research tools. RCaller brings all of the functionality that R serves in Java. Easy installation and integration steps and steep learning curve make RCaller a suitable solution. RCaller's computation overhead is generating XML files in R side and parsing XML to Java objects in Java side.
RCaller compiled jar library requires JRE (v11 or higher) and R installed in the runtime environment. If you want to compile from source, JDK (v11 or higher) and Maven (or Gradle) are also required for building process. Maven is responsible for downloading and install additional Java dependencies defined in pom file. RCaller does not use any version specific property of R.
It is recommended to take a look at the existing tests and examples
RCaller uses XML for reading R output by default. This is lightweight but not very fast way. For speeding up the IO, install R arrow package and add dependencies org.apache.arrow:arrow-vector and org.apache.arrow:arrow-memory-netty to your Java project. RCaller will use Arrow format automatically if it is available both in R and Java.
Errors handling can impact performance in online mode and is disabled by default.
Use RCaller.runAndReturnResultOnline(String var, boolean addTryCatch) method with addTryCatch = true for throwing R exceptions to Java.
Here is the auto-generated Javadocs.
There are many options to integrate RCaller with a Java Project. First option is to download pre-compiled jar file and add it to the classpath or pom.xml. pre-compiled jar files are here.
The second option is to compile from source. R should be installed to run tests. After building process, if everything is okay, a jar file is located in the target directory.
$ git clone https://github.com/jbytecode/rcaller.git
$ cd RCaller
$ mvn packageThe last option is to use maven dependency:
<dependency>
<groupId>com.github.jbytecode</groupId>
<artifactId>RCaller</artifactId>
<version>4.0.2</version>
</dependency>The RCaller is located in the Maven Central Repository.
You can test the performance of the library using the benchmark tool hosted here. This file tests costs of passing vectors and matrices from Java to R and vice versa.
If you want to join us, please follow the community guidelines here. We look forward to seeing your contributions.
Please refer to citation information given in file bibtex in your publications.